case also suggests that the random-effect model is the preferred way of analysis. So, from the
results of model 4, we can say that FDI and its higher inflow in the group of panel countries,
contribute to higher growth.
In addition, we have analysed another model in which random effect is present but we
have fixed period-specific effects, and results are reported under model 5. Model 5 reports
that exports and high level of FDI will increase the growth, otherwise FDI decreases growth
of the panel countries.
We also analyze the random-effect model by assuming the period-specific effect which
is also random (we call it the two-way random-effect model) and we report the results under
model 6. We find that in this case FDI, square of FDI, exports and labour force, are all found to
have positive impact on the economic growth in the panel of countries. Further, by providing
cross-section weights in the two-way random-effect model we find that results reported by
model 5 are robust to the inclusion of cross-section weights.
In the final step, in model 7, we used a random-effect model with the presence of a firstorder
autoregressive scheme. The results of model 7 reveal that higher inflows of FDI, exports,
and capital have positive and significant effect on the economic growth of our panel countries.
Further, we have proceeded to analyse the nonlinear impact of exports in the panel countries.
Results of nonlinear impact analysis of exports are presented in Table 2.
In Table 2, the results of the Hausman test show that the random-effect model is an
appropriate test for the analysis. The results of this model are reported under model 2. It is
evident from model 2 that FDI, exports, squared exports and labour force have positive and
significant impact on the economic growth of the panel countries. It also implies that when we
analyse the nonlinearity in both cases i.e., exports and FDI, we find a significant and positive
impact of exports only on the economic growth of panel countries. This also suggests the
preference of the export-led growth hypothesis against the FDI-led growth hypothesis (a long
debated topic) in our panel countries.7
Further, we have analyzed a model of random effect in which the period-specific effect is
assumed fixed and results are reported under model 3. We find very surprising results from
model 3. In this case, exports and FDI are significant with a negative coefficient, while the
coefficients of the square of exports and FDI are significant with a positive sign. Further, if
we compare the coefficient of exports and FDI, we find that the negative impact of FDI is
much higher with respect to the negative impact of exports. Similarly, the positive impact of
the square of FDI is also much higher vis-à-vis the positive impact of the square of exports.
In the final step, we have analysed a model of two-way random effect and results are
reported under model 4. The two-way random-effect model confirms the findings of the oneway
random-effect model, model 2; i.e., FDI, exports, squared exports and labour force have
positive and significant impact on the economic growth of the panel countries.